Ever since its discovery, the prominent UV feature in the interstellar extinction curve has been attributed to graphite - or at least some form of carbon-rich material - due to cosmic abundance constraints, and the fact that small graphite particles produce a feature in roughly the right wavelength range.
where x is the inverse wavelength (wavenumber in ), A is the magnitude of extinction, is close to the peak wavenumber of the UV feature, and are fitting parameters. The far-UV curvature is described by for and for . The function is referred to as a "Drude" profile, though it is a profile produced by small spheres described either by a single Lorentz or Drude oscillator (Bohren & Huffman 1983). This fit reproduces most if not all interstellar extinction curves known to date, with no systematic deviations or additional features or shoulders (at least, none beyond the observational errors).
Various lines of sight show peculiar extinction curves depending on the type of environment they pass through. For example, in the hydrogen poor circumstellar environment of R CrB, the bump is weaker and shifted to . Around carbon-rich (and hydrogen-rich) asymptotic giant branch stars, the bump is considerably weakened or absent (e.g., Snow et al. 1987). In H II regions, the relative strength of the bump (parameter ) is weaker than in most other environments (Jenniskens & Greenberg 1993). However, the range of bump widths is similar to that of the diffuse interstellar medium. Broader bumps in H II regions have their peak position shifted to smaller x. In other dense environments, like "bubbles" (regions showing loops and filaments characteristic of material swept up by stellar winds of OB stars in regions of recent massive star formation), the strength of the bump is similar to that of the diffuse ISM and no correlation is observed between its width and its peak position.
In this paper constraints on the optical properties of the purported interstellar UV feature carrier are derived considering rather general arguments related to chemical composition (as modelled by graphite or Lorentz oscillator models) and clustering (based on direct computations and interpretation using a spectral representation). In Sect. 2 some current theoretical models attempting to explain the characteristics of the interstellar UV feature are discussed. Their main shortcomings are emphasized. In Sect. 3 the profile of the UV feature of graphite is computed for various arrangements of touching spheres. The results are compared qualitatively to the observational constraints. Apart from the peak position falling at the wrong wavelength, the qualitative agreement is found to be good for compact clusters. But variations in chemical composition cannot be included directly in such a model. Therefore, in Sect. 4 a series of single-Lorentz oscillator models are used in conjunction with clustering to investigate the range of parameters (which simulate variations in chemical composition) consistent with the observational constraints. The clustering is modelled via a spectral representation formalism. In Sect. 5 the effects of adding a second Lorentz oscillator to explain the FUV rise are discussed. Interstellar curves along specific lines of sight exhibiting a wide range of FUV curvatures are modelled and plausible (though non-unique) optical constants for the bump grains along these lines of sight are derived. In Sect. 6 physical mechanisms relevant to the expected optical properties of the UV feature carrier are discussed, in particular, dehydrogenation and UV processing.
© European Southern Observatory (ESO) 1997
Online publication: June 5, 1998